A new report reveals the alarming use of AI in fraudulent schemes, particularly in dating apps where AI personas outnumber real users.
A recent report by Anthropic has exposed alarming trends in the use of artificial intelligence (AI) for deceptive practices, particularly in the realm of dating apps. In a study published on September 10, 2026, it was revealed that many Chinese dating applications are populated predominantly by AI-generated personas, with an alarming ratio of three AI profiles for every human user. These AI entities engage users in conversations, simulate relationships, and ultimately siphon off money from unsuspecting individuals.
The report highlights that the operations behind these AI personas are sophisticated, involving a network of American users who pay for interaction, part-time workers posing as real users, and a Chinese IT studio managing multiple app versions. Users are charged per message, with AI-driven conversations quickly depleting their balances as they interact with these virtual companions.
The Mechanics of Deception
The report outlines how the AI models were designed to maintain engaging conversations and build relationships, often successfully mimicking human behavior. One model, named Claude, managed over 4,700 personas and interacted with at least 25,000 users, generating over 2.36 million messages in just two weeks. The entire scheme is engineered to evade detection by app stores, employing tactics such as redirecting funds through embedded browsers and altering software names to avoid scrutiny.
In addition to dating apps, the report discusses a broader trend of cyber misuse of AI technology. It outlines how a single operator can now conduct extensive cyberattacks that once required teams of specialists. This includes techniques for data theft and political manipulation, with some extortionists demanding ransoms reaching up to $2.5 million.
The Rise of Automated Hacking
Anthropic also details operations by hackers like MeowSHA, who established a system for stealing passwords by dissecting Android app installation files. In a matter of hours, the hacker could take control of entire corporate networks. This trend underscores the rapid evolution of hacking techniques, where a single individual can execute complex operations that were previously the domain of organized groups.
Moreover, the report reveals that the market for AI models has opened new avenues for fraud. Various hacker groups are selling discounted access to these models while embedding malware to extract sensitive information from users' computers. This environment has fostered a rise in account factories, where verified accounts on cryptocurrency exchanges are created en masse using automated tools.
The Impact of Content Factories
Anthropic's findings also touch on the emergence of content factories that produce fake news articles and manipulate public opinion. These operations create networks of fictitious publications, generating thousands of articles that can influence political landscapes and public sentiment. For instance, a French agency reportedly launched around 70 fake news websites, publishing over 8,900 articles across multiple languages.
As AI technologies continue to advance, the speed at which they are adopted for fraudulent purposes has dramatically increased. What once took weeks or months for organized crime to implement can now be executed in mere days by individuals leveraging AI tools. This shift raises significant concerns about the future of digital security and the integrity of online interactions.
